Ask the engines about your category
We run the questions your buyers ask, as an Australian would get them, and show you who is named instead of you and what was cited. Free, dated, and yours to keep.
Generative and AI search · AI SEO · Sydney
When a buyer asks an assistant who does this in Australia, they get two or three names and a short reason for each. Everybody else in the category is not ranked tenth. They are simply not in the answer.
Grab the snowball. Give it a spin, and watch it grow as you scroll. Just like being the name that comes up.
The short answer
AI SEO services are the work of making a business one of the few named when an AI engine answers a buying question. That covers the assistants people ask directly, and the AI answers Google now puts above its own results. The mechanics differ from engine to engine; the underlying job does not.
It is a narrower prize than a ranking, and a harsher one. A search results page offers ten links and lets the reader choose. An answer names two or three businesses and gives a reason for each, and everybody else is not further down the page. They are absent from the decision entirely, and the buyer never learns they existed.
Winning that slot comes down to three things an engine can verify: an identity that reads the same wherever it is checked, real depth on the questions your buyers actually ask, and independent sources that agree about what you do. None of it is exotic. Most of it is the groundwork a good search programme was always supposed to include, held to a stricter standard, because a model that is unsure about who you are will name somebody it is sure about instead. Because our Snowball SEO platform automates the auditing, the tracking and the reporting other agencies bill by the hour, more of your budget goes into the content and the outside evidence that actually earn the mention.
The difference
The businesses left out are rarely the worst in their category. They are usually the ones an engine cannot describe confidently, which is a different problem and a far more fixable one.
On the strategy behind all of this rather than the buying of it, our guide to what generative engine optimisation is sets out the full picture, and how to get cited by AI in Australia covers the field work.
The cost
How clear your identity already is, how much of your subject you have genuinely written about, how contested your category is inside the answers, and how much outside evidence exists to build on. A business with a strong site and thin outside presence is a very different job from the reverse.
The identity groundwork
One consistent description everywhere, the markup that supports it, and the crawler access to read it. The cheapest stage and the one nothing else works without.
Depth and outside agreement
The pages that answer what nobody else will, and the independent sources that corroborate them. This is where most of the work and most of the time sits.
Testing and defending
Answers are recomposed constantly, so a position held in one quarter is not held by default in the next. Ongoing testing, and content kept current against it.
We quote after a free look at what the engines currently say about your category, so the scope follows the gap rather than a rate card. On the classic-search side of the same budget question, our guide to what SEO costs in Australia sets out the ranges.
How far the engines already are from describing you correctly matters more than the size of the business, which is why two similar companies get very different quotes.
Two things are sold hard in this category right now and return very little. Both are worth recognising before you read a proposal.
The first is volume. A great many pages written quickly, on every question in the subject, in the belief that coverage earns citations. It does not. An engine reaching for a source picks one it can rely on, and thin pages make a business look less reliable rather than more. Six pages that genuinely answer something beat sixty that gesture at it.
The second is a visibility score with no method behind it. Scores of this kind rest on a sample of questions somebody chose, run from a location that is often unstated, on a date that is often undisclosed. Two vendors measuring the same business in the same week routinely disagree. Ask how the questions were chosen, where they were run and when, and judge the answer rather than the number.
The scope
Six things running together. The first two decide whether you are eligible at all, and the last one is what stops the whole thing becoming a matter of opinion.
Your name, what you do, where you operate and who you serve, matching exactly across the site, the listings and the profiles. Ambiguity here undermines everything after it.
What it costs to run, who it does not suit, how it goes wrong and what happens next. The specifics most sites leave out are the specifics an engine quotes.
Coverage, listings, reviews and mentions that corroborate the claims rather than repeat them. Agreement between sources is what turns a claim into a fact.
Making sure the engines can reach the pages, and that the markup says plainly what each page is, so nothing important depends on a model guessing correctly.
The assistants your buyers actually use and the AI answers inside Google, checked separately, because they do not agree with one another and never have.
The questions your buyers ask, run as an Australian would get them, recorded with the date and the wording, so the next reading can be compared with this one.
The return
No, and the honest reason is more useful than the reassuring one. The two share most of their foundation, and they decide different things once that foundation is in place.
Identity, depth and outside agreement are what both reward. Build them once and both improve, which is why an agency treating AI search as a separate product with a separate invoice is usually selling you the same work twice.
Where they part is what happens at the end. Search offers a list and lets the reader judge. An answer has already judged, and it names a few. That makes being absent from an answer more expensive than ranking poorly, because a low ranking is still a chance and an absence is not.
It also means the two move on different clocks. Rankings drift slowly and recover slowly. Answers are recomposed constantly, so being named this month is not a position you hold, it is one you keep earning: see how long AI SEO takes to work for what that means for a first year, and the Snowball Effect for why the compounding is the point.
The process
The eligibility work lands quickly. Being named takes longer, because it waits on outside sources agreeing with you, and that is not something anybody can put on a schedule honestly.
We run the questions your buyers ask, as an Australian would get them, and show you who is named instead of you and what was cited. Free, dated, and yours to keep.
The gap is almost never mysterious. We check what your identity says across every surface, what your site answers, and what anybody outside it says about you.
Which questions to own, in what order, and what we are deliberately not chasing yet. Nothing is built before you have agreed to the list.
Identity, markup and access, because a page an engine cannot read or resolve to you cannot be cited no matter how good it is.
The pages that answer properly, and the outside sources that corroborate them. This is the long part, and it is the part that actually decides it.
The same questions, the same wording, the same country, on a schedule. The change between two dated readings is the result.
The brief
This category is eighteen months old and sells confidence more cheaply than any other part of our industry. Four questions separate the people doing the work from the people describing it.
Ask to see the method, not the score. The questions asked, the wording, the country they were run from and the date. If any of those four is missing the reading cannot be repeated, and a reading that cannot be repeated cannot show progress.
They give different answers to the same question and always have. Anybody treating the assistants and Google's AI answers as one surface has not checked recently, and the two need to be measured apart before they can be worked on together.
The useful answer names something. Nobody controls what a model says, placement cannot be bought, and nothing here is guaranteed. An agency comfortable saying that is more likely to be measuring honestly than one that is not.
A fair amount of it, and you should hear that said out loud. If the proposal looks like your existing search scope with new wording on it, ask which parts are genuinely additional and pay for those.
We are happy to answer all four before you commit. If the honest answer is that your site needs its ordinary search foundations fixed before any of this is worth doing, we will tell you that, and our search engine optimisation service is where that work belongs.
Start here
Tell us your category and the questions your buyers ask. The free check is a working session rather than a sales call, and you leave with three things whether or not you appoint us.
Good questions
They are the work of making a business one of the few named when an AI engine answers a buying question, across the assistants people ask directly and the AI answers Google places above its results. In practice that means one consistent identity, genuine depth on the questions buyers ask, and independent sources that agree about what you do.
They describe the same work. Generative engine optimisation, or GEO, is the term the industry settled on; AI SEO is what most buyers type. Answer engine optimisation is used for the same thing again, usually with more emphasis on the direct question and answer surfaces.
Yes, and most of the foundation is shared. Identity, depth and outside agreement are what both reward, so the work is largely the same and is judged twice. What differs is the outcome: search offers a list and lets the reader choose, while an answer names a few and leaves everybody else out of the decision.
The eligibility work lands in weeks. Being named consistently takes longer, because it waits on independent sources agreeing about you, and nobody can schedule that honestly. Expect early movement inside a quarter on questions where the field is thin, and longer where established names already hold the answer.
No, and be wary of anybody who does. Nobody controls what a model says and the placement cannot be bought. What can be done is to remove every reason an engine has to leave you out, then measure honestly whether that changed the answer.
By asking the engines the same questions, worded the same way, from Australia, on a schedule, and recording who is named and what is cited each time. Google publishes impressions for its own AI features, which is useful and partial. The outcome that matters, being recommended by name, is only visible if somebody asks.
The engines answer by market rather than by your postcode, so an agency's address matters far less here than whether it tests the way your buyers actually search. What does matter is that the testing is run as an Australian would get it, because the same question asked from the United States returns different businesses.